6/5/2017 HashingConceptsinDBMS

Introduction

HashFileorganizationmethodistheonewheredataisstoredatthedatablockswhoseaddressisgeneratedby
usinghashfunction.Thememorylocationwheretheserecordsarestorediscalledasdatablockordatabucket.
Thisdatabucketiscapableofstoringoneormorerecords.

Thehashfunctioncanuseanyofthecolumnvaluetogeneratetheaddress.Mostofthetime,hashfunctionuses
primarykeytogeneratethehashindexaddressofthedatablock.Hashfunctioncanbesimplemathematical
functiontoanycomplexmathematicalfunction.Wecanevenconsiderprimarykeyitselfasaddressofthedata
block.Thatmeanseachrowwillbestoredatthedatablockwhoseaddresswillbesameasprimarykey.This
implieshowsimpleahashfunctioncanbeindatabase.

Abovediagramdepictsdatablockaddresssameasprimarykeyvalue.Thishashfunctioncanalsobesimple
mathematicalfunctionlikemod,sin,cos,exponentialetc.Imaginewehavehashfunctionasmod(5)todetermine
theaddressofthedatablock.Sowhathappenstotheabovecase?Itappliesmod(5)onprimarykeysand
generates3,3,1,4and2respectivelyandtherecordsarestoredinthosedatablockaddresses.

https://www.tutorialcup.com/dbms/hashingconcepts.htm 1/7
6/5/2017 HashingConceptsinDBMS

Fromabovetwodiagramsitnowclearhowhashfunctionworks.

TherearetwotypesofhashfileorganizationsStaticandDynamicHashing.

StaticHashing

Inthismethodofhashing,theresultantdatabucketaddresswillbealwayssame.Thatmeans,ifwewantto
generateaddressforEMP_ID=103usingmod(5)hashfunction,italwaysresultinthesamebucketaddress3.
Therewillnotbeanychangestothebucketaddresshere.Hencenumberofdatabucketsinthememoryforthis
statichashingremainsconstantthroughout.Inourexample,wewillhavefivedatabucketsinthememoryusedto
storethedata.

Searchingarecord
https://www.tutorialcup.com/dbms/hashingconcepts.htm 2/7
6/5/2017 HashingConceptsinDBMS

Usingthehashfunction,databucketaddressisgeneratedforthehashkey.Therecordisthenretrievedfromthat
location.i.e.ifwewanttoretrievewholerecordforID104,andifthehashfunctionismod(5)onID,theaddress
generatedwouldbe4.Thenwewilldirectlygottoaddress4andretrievethewholerecordforID104.HereID
actsasahashkey.

Insertingarecord

Whenanewrecordneedstobeinsertedintothetable,wewillgenerateaaddressforthenewrecordbasedonits
hashkey.Oncetheaddressisgenerated,therecordisstoredinthatlocation.

Deletearecord

Usingthehashfunctionwewillfirstfetchtherecordwhichissupposedtobedeleted.Thenwewillremovethe
recordsforthataddressinmemory.

Updatearecord

Datarecordmarkedforupdatewillbesearchedusingstatichashfunctionandthenrecordinthataddressis
updated.

Supposewehavetoinsertsomerecordsintothefile.Butthedatabucketaddressgeneratedbythehashfunction
isfullorthedataalreadyexistsinthataddress.Howdoweinsertthedata?Thissituationinthestatichashingis
calledbucketoverflow.Thisisoneofthecriticalsituations/drawbackinthismethod.Wherewillwesavethedata
inthiscase?Wecannotlosethedata.Therearevariousmethodstoovercomethissituation.Mostcommonly
usedmethodsarelistedbelow:

Closedhashing

Inthismethodweintroduceanewdatabucketwithsameaddressandlinkitafterthefulldatabucket.These
methodsofovercomingthebucketoverflowarecalledclosedhashingoroverflowchaining.

ConsiderwehavetoinsertanewrecordR2intothetables.Thestatichashfunctiongeneratesthedatabucket
addressasAACDBF.Butthisbucketisfulltostorethenewdata.Whatisdoneinthiscaseisanewdatabucket
isaddedattheendofAACDBFdatabucketandislinkedtoit.ThennewrecordR2isinsertedintothenew
bucket.Thusitmaintainsthestatichashingaddress.Itcanaddanynumberofnewdatabuckets,whenitisfull.

https://www.tutorialcup.com/dbms/hashingconcepts.htm 3/7
6/5/2017 HashingConceptsinDBMS

OpenHashing

Inthismethod,nextavailabledatablockisusedtoenterthenewrecord,insteadofoverwritingontheolderone.
ThismethodiscalledOpenHashingorlinearprobing.

Inthebelowexample,R2isanewrecordwhichneedstobeinserted.Butthehashfunctiongeneratesaddressas
237.Butitisalreadyfull.Sothesystemsearchesnextavailabledatabucket,238andassignsR2toit.

Inthelinearprobing,thedifferencebetweentheolderbucketandthenewbucketisusuallyfixedanditwillbe1
mostofthecases.

Quadraticprobing

Thisissimilartolinearprobing.Buthere,thedifferencebetweenoldandnewbucketislinear.Weusequadratic
functiontodeterminethenewbucketaddress.

https://www.tutorialcup.com/dbms/hashingconcepts.htm 4/7
6/5/2017 HashingConceptsinDBMS

DoubleHashing

Thisisalsoanothermethodoflinearprobing.Herethedifferenceisfixedlikeinlinearprobing,butthisfixed
differenceiscalculatedbyusinganotherhashfunction.Hencethenameisdoublehashing.

DynamicHashing

Thishashingmethodisusedtoovercometheproblemsofstatichashingbucketoverflow.Inthismethodof
hashing,databucketsgrowsorshrinksastherecordsincreasesordecreases.Thismethodofhashingisalso
knownasextendablehashingmethod.Letusseeanexampletounderstandthismethod.

ConsidertherearethreerecordsR1,R2andR4areinthetable.Theserecordsgenerateaddresses100100,
010110and110110respectively.Thismethodofstoringconsidersonlypartofthisaddressespeciallyonlyfirst
onebittostorethedata.Soittriestoloadthreeofthemataddress0and1.

WhatwillhappentoR3here?ThereisnobucketspaceforR3.Thebuckethastogrowdynamicallyto
accommodateR3.Soitchangestheaddresshave2bitsratherthan1bit,andthenitupdatestheexistingdatato
have2bitaddress.ThenittriestoaccommodateR3.

NowwecanseethataddressofR1andR2arechangedtoreflectthenewaddressandR3isalsoinserted.As
thesizeofthedataincreases,ittriestoinsertintheexistingbuckets.Ifnobucketsareavailable,thenumberof
bitsisincreasedtoconsiderlargeraddress,andhenceincreasingthebuckets.Ifwedeleteanyrecordandifthe
datascanbestoredwithlesserbuckets,itshrinksthebucketsize.Itdoestheoppositeofwhatwehaveseen
above.Thisishowadynamichashingworks.Initiallyonlypartialindex/addressgeneratedbythehashfunctionis
consideredtostorethedata.Asthenumberofdataincreasesandthereisaneedformorebucket,largerpartof
theindexisconsidertostorethedata.

AdvantagesofDynamichashing
https://www.tutorialcup.com/dbms/hashingconcepts.htm 5/7
6/5/2017 HashingConceptsinDBMS

Performancedoesnotcomedownasthedatagrowsinthesystem.Itsimplyincreasesthememorysizeto
accommodatethedata.
Sinceitgrowsandshrinkswiththedata,memoryiswellutilized.Therewillnotbeanyunusedmemorylying.
Goodfordynamicdatabaseswheredatagrowsandshrinksfrequently.

DisadvantagesofDynamichashing

Asthedatasizeincreases,thebucketsizeisalsoincreased.Theseaddresseswillbemaintainedinbucket
addresstables.Thisisbecause,theaddressofthedatawillkeepchangingasbucketsgrowandshrink.When
thereisahugeincreaseindata,maintainingthisbucketaddresstablebecomestedious.
Bucketoverflowsituationwilloccurinthiscasetoo.Butitmighttakelittletimetoreachthissituationthanstatic
hashing.

ComparisonofOrderedIndexingandHashing

OrderedIndexing Hashing

Addressesinthememoryaresortedforkey Addressesaregeneratedusinghashfunctionon
value.Thiskeyvaluecanbeprimarykeyorany thekeyvalue.Thiskeyvaluecanbeprimarykey
othercolumninthetable. oranyothercolumninthetable.

Performanceofthismethodcomesdownasthe Performanceofdynamichashingwillbegood
dataincreasesinthefile.Sinceitstoresthedata whenthereisafrequentadditionanddeletionof
inasortedform,whenthereis data.Butifthedatabaseisveryhuge,
insert/delete/updateoperation,anextraeffortto maintenancewillbecostlier.
sorttherecordisneeded.Thisreducesits
performance.
Statichashingwillbegoodforsmaller
databaseswhererecordsizeidpreviously
known.Ifthereisagrowthindata,itresultsin
seriousproblemslikebucketoverflow.

https://www.tutorialcup.com/dbms/hashingconcepts.htm 6/7
6/5/2017 HashingConceptsinDBMS

Therewillbeunuseddatablocksdueto Inbothstaticanddynamichashing,memoryis
delete/updateoperation.Thesedatablockswill wellmanaged.Bucketoverflowisalsohandled
notbereleasedforreuse.Henceperiodic tobetterextentinstatichashing.Datablocks
maintenanceofthememoryisrequired.Else, aredesignedtoshrinkandgrowindynamic
memoryiswastedandperformancewillalso hashing.
degrade.Alsoitwillbecostoverheadto
maintainmemory.
Buttherewillbeanoverheadofmaintainingthe
bucketaddresstableindynamichashingwhen
thereisahugedatabasegrowth.

Preferredforrangeretrievalofdatathatmeans Thismethodissuitabletoretrieveaparticular
whenthereisretrievaldataforparticularrange, recordbasedonthesearchkey.Butitwillnot
thismethodisbestsuited. performbetterifthehashfunctionisnotonthe
searchkey.

https://www.tutorialcup.com/dbms/hashingconcepts.htm 7/7